X3T9.2/92-158 R0 To: Membership of X3T9.2 From: Lawrence J. Lamers, Secretary X3T9.2 John B. Lohmeyer, Chair X3T9.2 Date: July 23, 1992 Subject: Minutes of X3T9.2 July '92 General Working Group ______________________________________________________________________________ Agenda 1. Opening Remarks 2. Attendance and Membership 3. Draft ATA Public Review responses [Allan] 4. ATA Extensions - DMA Timing [Emulex] 5. Revised SCSI-3 command set project proposals (92-106,107,108,109) [Lohmeyer/Lamers] 6. Draft CAM-2 project proposal [Gallant/Allan] 7. ASCQ's for Data Compression (92-134) [Williams] 8. Proposal to add a disconnect unconditionally after command phase (92-133) [Smyers] 9. Blind pluggable connector (92-139) [Cornaby] 10. SCSI-3 I/O Process Queuing (91-98R9) [Penokie] 11. SCSI-2 TIB on I/O Process Queuing [Gallant] 12. Synchronization Status and Errors (92-130) [Robison] 13. Enhancement of Sequential Access for Medium Changers (92-006) [Jones] 14. Capacitance on SCSI Signal Wires (92-143) [McGrath] 15. Change Mode Page (92-092R2) [Penokie] 16. SCSI-3 Architecture Model (92-079) [Monia] 17. Serial Bus Protocol (92-135) [Gardner] 18. Directly Addressable Device Interface [Smyers] 19. Fiber Channel Protocol (91-122) [Snively] 20. Generic Packetized Protocol (91-013) [Stephens] 21. Working Group Schedule 22. Adjournment Results of Meeting 1. Opening Remarks Dal Allan, the Vice Chair, called the meeting to order at 2:00 p.m., Monday July 20, 1992 for the session devoted to ATA comment resolution and enhanced ATA. John Lohmeyer chaired the Tuesday morning and Wednesday sessions. They both thanked Horst Truested of IBM for hosting the meeting. As is customary, the people attending introduced themselves. A copy of the attendance list was circulated for attendance and corrections. It was stated that the meeting had been authorized by X3T9.2 and would be conducted under the X3 rules. Ad hoc meetings take no final actions, but prepare recommendations for approval by the X3T9.2 committee on Lower Level Interfaces. The voting rules for the meeting are those of the parent committee, X3T9.2. These rules are: one vote per company; and any participating company member may vote. The minutes of this meeting will be posted to the SCSI BBS and the SCSI Reflector within 14 calendar days and will be included in the next committee mailing. 2. Attendance and Membership Attendance at working group meetings does not count toward minimum attendance requirements for X3T9.2 membership. Working group meetings are open to any person or company to attend and to express their opinion on the subjects being discussed. The following people attended the meeting: Name Status Organization ------------------------------ ------ ------------------------------ Mr. Wayne Sanderson O Mr. Norm Harris P Adaptec, Inc. Mr. Charles Brill P AMP, Inc. Mr. Bob Whiteman A AMP, Inc. Mr. Bill Garver O AMP, Inc. Mr. Michael Wingard A Amphenol Interconnect Mr. Jan V. Dedek P Ancot Corp. Mr. Scott Smyers P Apple Computer Mr. Robert Brown O Areal Technology Mr. Bob Gannon O C&M Corp. Mr. David Ruxton O Calluna Technology Limited Mr. James Woo V Cirrus Logic Inc. Mr. Stephen R. Cornaby P Conner Peripherals Mr. George Zisk O CW Industries Mr. Charles Monia P Digital Equipment Corp. Dr. William Ham A Digital Equipment Corp. Mr. John A. Gallant A Digital Equipment Corp. Mr. Edward A. Gardner A Digital Equipment Corp. Mr. Ralph Weber A Digital Equipment Corp. Mr. Paul R. Nitza A Emulex Corp. Mr. I. Dal Allan P ENDL Mr. D. W. Spence A ENDL Associates Mr. Robert Liu P Fujitsu Computer Products,Am Mr. Jeffrey L. Williams P Hewlett Packard Co. Mr. Howard Wang O Hitachi Computer Products Mr. Wayne Kosters V I-TECH Corp. Mr. Tsuyoshi Takizawa O IBM (Japan) Mr. George Penokie P IBM Corp. Mr. Gary R. Stephens A IBM Corp. Mr. Otis Greene O IBM Corp. Mr. Jonathan Fasig V IBM Corp. Mr. John P. Scheible V IBM Corp. Mr. John Blagaila S Integral Peripherals Mr. Kevin R. Pokorney O Intellistor, Inc. Mr. Lawrence J. Lamers P Maxtor Corp. Mr. Raymond C. Yule P Micropolis Corp. Mr. Reginald J. Murray S Ministor Peripherals Corp. Mr. Robbie Shergill P National Semiconductor Mr. John Lohmeyer P NCR Corp. Mr. Rod DeKoning V NCR Corp. Mr. Stephen F. Heil P Panasonic Technologies, Inc Mr. James McGrath P Quantum Corp. Mr. Hale Landis A Seagate Technology Mr. Ram Battu S Seagate Technology, Inc. Mr. Daniel E. Moczarny O Silicon Systems, Inc. Mr. Robert N. Snively P Sun Microsystems, Inc. Mr. Harvey Waltersdorf P Thomas & Betts Mr. Shishir Shah A Western Digital Mr. Tak Asami O Western Digital 49 People Present Status Key: P - Principal Member A - Alternate Member L - Liaison Member O - Observer S,V - Visitor 3. Draft ATA Public Review responses [Allan] There was a discussion on CORR, UNC, and ERR bits as they relate to READ LONG. The ERR bit will be put back as a 'V'. Jim McGrath's comment on 3 volt operation and download microcode command was discussed. The group requested Jim to come up with a proposal for the download microcode command. Shishir's comment on Logical Block Addressing and greater than 528 MByte addressing was discussed. Shishir had the proposed wording for the LBA issue. This had been discussed at previous meetings. Power pins - We had expected a comment on power pins in Saied's public review comment, but it was not there. Hal Landis had not yet put together an industry opinion on this. 4. ATA Extensions - DMA Timing [Emulex] Mehran Ramezani presented a proposal that combined input from WD and Cirrus Logic for a Mode 3 transfer. There was a lengthy discussion on setup time. The result was a footnote specifying a setup time as the difference between tD min and tE max. Is there a need for single and multiple transfer? Does Mode 3 mean only Mode 3 or Mode 0 to 3. This is a software problem that needs a bit significant indication of the modes supported. Cirrus Logic made a proposal for this. A new word was proposed for reporting DMA, with the lower 8 bits reporting capability and the upper 8 bits reporting current setting. The revised proposal will be developed by Cirrus Logic. 5. Revised SCSI-3 command set project proposals (92-106,107,108,109) [Lohmeyer/Lamers] There was a discussion on how to handle the SCSI core commands. The working group favored having a separate document. John Lohmeyer mentioned the movement toward electronic distribution and notification for committee business. He stated he will be pursuing this topic with X3. X3 is currently offering a service called OASIS on Compuserve. Further information on this service will be in the next mailing. 6. Draft CAM-2 project proposal [Gallant/Allan] John Gallant presented a project proposal for an extended common access method that would provide for 64-bit addressing. There were several inputs from the working group which John will incorporate in the proposal. 7. ASCQ's for Data Compression (92-134) [Williams] Jeff asked that this item be Deferred to the next working group meeting. 8. Proposal to add a disconnect unconditionally after command phase (92-133) [Smyers] Scott Smyers presented a proposal to add a disconnect immediate bit to mode page 2h. There was general consensus that this could be a useful feature. The bit will be moved to bit 3 of the byte. The working group recommends that this proposal be accepted for SCSI-3 as revised. 9. Blind pluggable connector (92-139) [Cornaby] Steve Cornaby discussed the pin assignments for the unitized connector. He is no longer referring to it as blind-pluggable because that is a mating connector issue and not germane to the discussion. Ed Gardner suggested a alternate pin assignment based on functional needs for RAIDs. There was agreement that the functions Ed enumerated need to be accommodated. John Lohmeyer brought up the question of which group should standardize the connector. The working group consensus was that the functions needed should be defined by the X3T9.2 working group, but all the mechanical and electrical issues should be addressed by the SFF committee. 10. SCSI-3 I/O Process Queuing (91-98R9) [Penokie] Deferred due to lack of time. 11. SCSI-2 TIB on I/O Process Queuing [Gallant] Deferred at John's request. 12. Synchronization Status and Errors (92-130) [Robison] Rod DeKoning presented a proposal to add sense codes to report synchronization information. Also a synchronization status field is proposed for the Rigid Disk Drive Geometry page. The working group liked adding the information, but did not like the proposed location for the information. It was suggested a way be found to put the status information in REQUEST SENSE data. Moving the RPL stuff out of the Rigid Disk Drive Geometry page into the Control Mode page was also suggested. A revised document will be created and distributed via the SCSI reflector for comment. 13. Enhancement of Sequential Access for Medium Changers (92-006) [Jones] Kurt Chan brought a revised proposal for consideration. No action was taken. 14. Capacitance on SCSI Signal Wires (92-143) [McGrath] This item was moved to the SPI agenda. 15. Change Mode Page (92-092R2) [Penokie] This proposal is the response to the rejection at the last meeting of using a command to change the mode of operation (i.e., the INQUIRY data). Paul Nitza objected to the ability to modify the INQUIRY data piecemeal. John Lohmeyer did not like this either and wanted some time to review the proposal. Getting back to the default parameters was raised as an issue. One possible solution to the restoration of default parameters is to add a Restore-Default- Parameters bit to the MODE SELECT command. This could be a convenient way to restore default parameters for any mode page. The topic needs further discussion. 16. SCSI-3 Architecture Model (92-079) [Monia] Charles had to wear his asbestos suit as he led a discussion of the draft SAM document. There was a considerable debate about the domain and the addressing. The meaning of LUN was also debated ad nauseam. There are two viewpoints that are architecturally different. One view states that the LUN is always the same regardless of the path to the device (multi-ported device). The other view is pure virtual reality, where the LUN is part of an address but not fixed on the device (LUN 2 on port 0 might be LUN 3 on port 1). The architecture for linked commands needs to be modified so that the intermediate status is returned for each command to advance the CDB pointer. This will allow the application to know which command terminated if there is an exception condition. A method of extending linked commands across packets needs to be established. 17. Serial Bus Protocol (92-135) [Gardner] The following are highlights of Ed Gardner's presentation on serial bus protocol: The P1394 project is attempting to integrate host adapter functions into the device and thereby remove one bottleneck in the I/O subsystem. A function is any request an initiator may give to a target. An FCB is a data structure that describes a function. LUN: a 32-bit identifier within the Control Status Register (CSR) address. The identifier needs to be assigned by an authority. The ID is in a ROM. initiator: multiple target CSRs or distinct initiator IDs within same target CSR. queue tag: 64-bit i/o process identifier maybe used as an "address" to access the command's FCB, e.g., to store autosense data. FCB delivery & completion: in proposal - write FCB to target CSR think of CSR as entry to FIFO "message passing" - write completion indication to initiator specified address might be interrupt FIFO latest thinking - mailbox scheme similar to Adaptec SCSI adapters At this point a general discussion/tutorial on mailbox approaches took place. Bob Snively gave the mailbox tutorial. Ed returned to his presentation: data transfer identifier: 16 bytes describes how to perform a command's data transfer. (8 bytes is 64-bit CSR address space) multiple formats: - contiguous: "address", length, flags (seq vs reordered) - scatter gather: "address" & length refer to list of contiguous pointers - isochronous: channel number, ??? Several questions arose during the discussion: Who is the owner of control - initializing device and who is permitted to give commands to device. What is the base address of CSR space? Dealing with dual-ported devices. If there are two independent CSR spaces, a device thinks two separate initiators, not one initiator with two access paths to same host. What happens if an initiator re-uses a queue tag when there is an active process in that address. The semantics on "globally accessible" memory remains an issue. 18. Directly Addressable Device Interface [Smyers] Scott Smyers raised a question on doing atomic operations for command delivery. Several system busses have difficulty with atomic operations (e.g., Nu-bus and micro-channel). A straw poll was taken of folks that thought atomic operations for command delivery would be acceptable. The result was everyone opposed use of atomic operations. Scott also got a ration of "input" regarding his intention not to use SCSI commands. 19. Fiber Channel Protocol (91-122) [Snively] Bob Snively presented an introduction to the FCP. There was some ensuing confusion because most of the SCSI folks did not have the Fibre Channel background to follow it. Bob promised a tutorial at the September X3T9.2 working group. 20. Generic Packetized Protocol (91-013) [Stephens] Gary plans to bring in a GPP for SPI. 21. Working Group Schedule The General Working Group meeting schedule is: Date Location Host ------------------ ----------------- ------------------------ Sep 21-25, 1992 Toronto, Canada Canstar Nov 9-13, 1992 Sunnyvale, CA Amdahl Jan 11-15, 1993 Sunnyvale, CA Tandem Mar 15-19, 1993 Orange County, CA Emulex May 17-21, 1993 Sante Fe, NM Los Alamos Jul 19-23, 1993 Manchester, NH DEC Sep 13-17, 1993 Poughkeepsie, NY IBM Nov 8-12, 1993 Colo Spgs, CO NCR The tentative meeting schedule for the X3T9.2 September working group week is: Day Meeting Times --------- ---------------------------------- -------------- Monday Small Form Factor Committee 9:00a - 1:00p ATA Extensions 2:00p - 5:00p Tuesday X3T9.2 General Working Group 9:00a - 1:00p SPI Working Group 2:00p - 8:00p Wednesday X3T9.2 General Working Group 9:00a - 5:00p Thursday SCSI Editors 9:00a - 5:00p 22. Adjournment The meeting was adjourned at 6:00 p.m. on Wednesday July 22, 1992.